Analysis About Surface Texture Of Three-dimensional Weaving Piece Based On Wavelet Transformation

The three dimension-weaving piece studied in the thesis is a new, advanced enhanced fabric and it is necessary for advanced and multi-function composite form. It is a new material needed badly in fields of aviation and space flight. The three dimension weaving technology of the composite is a new and high technology developed in eighties of the 20th century. The global can of it is good and the structure of it is designed flexibly. So this composite receives much concern at home and abroad. It's two parameters: the weaving angle and the length of the texture are the key of monitoring and controlling the quality of three dimension weaving piece. So it is very meaningful to determine the two parameters. The subject is mainly focused on the weaving angle's determination, studies the surface image processing system and the surface image testing system and make some achievements in research. 1. Directed against the three dimension weaving piece's image's lots of noise and its own texture properties, The image strengthen technology on the basis of the wavelet transforms is used to realize the edge's drawing of the three dimension weaving piece's surface image. The wavelet's time, frequency characteristic combined with the unsharp masking is used to enhance the surface image of the three dimension-weaving piece and the edge can be drew. The better result can be obtained compared with the traditional edge drawing operators( Roberts operator, Sobel operator, Prewitt operator, Krisch operator, Gauss-Laplas operator). 2. The part-threshold selecting algorithm is improved by use of the inside difference of the image. It is used in surface image segmenting of the three dimension-weaving piece and good segmenting result can be obtained. It can be proved by the research that using this algorithm can avoid the reflecting of different even degree of illumination.3. The idea of linear fitting is applied to the unit weaving angle's computation of the three dimension-weaving piece. Linear fitting method makes full use of the histogram information of all pixels in a texture unit and good result can be obtained through this method.4. Focusing on the self-organization artificial neural network, the basic competing neural network can be expended, which makes the mode that wins victory gained through the neural network is the dynamic optimization in the direction of some mode. So better computation result of the weaving angle can be obtained.5. The traditional template matching method is improved to define the position of every texture unit in a texture image. Combined this improved method with the linear fitting method and the self-organization artificial neural network method, the average weaving angle can be obtained and at the same time the irregular degree of the weaving angle can be evaluated. And then more overall technical support can be offered to the making of the three dimensional Weaving Piece.